Transductive Support Vector Machines (TSVM) With Hybrid Recurrent Neural Networks For Medical Image Filtering

Abstract

Filtering an image is one of the most fundamental and consequential tasks in the field of image processing. As of now, scientists still face a very real and pressing dilemma. Images, whether in print or digital form, are a crucial criterion in many different industries, including geosciences, aerospace, education, agriculture, entertainment, surveillance, and many more. Background noise degrades image quality, despite extensive research aimed at finding a solution to this problem and the anticipation of a plethora of possible countermeasures. Methods for picture filtering, including those based on machine learning and those based on optimization, are detailed here. It has been possible to categorize these methods by the methods they employ. Gaussian noise, white noise, salt and pepper noise, speckle noise, and poison noise were all removed from lung CT scans using a combination of Recurrent Neural Networks (RNNs) and Transductive Support Vector Machines (TSVMs) in this study. Using TSVM, RNN can function better than ever before. The approach proposed here uses LSTM to normalize batches of data. By simulating a large number of individual beehives, the ABC (Artificial Bee Colony) algorithm may determine the optimal size of each batch. MATLAB was used to realize the suggested method. The experimental consequences demonstrate that the suggested system overtakes the baseline in terms of PSNR, MSE, and precision.